2019.12.23.887182V1.Full.Pdf

bioRxiv preprint doi: https://doi.org/10.1101/2019.12.23.887182; this version posted December 23, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

1 Two distinct bacterial biofilm components trigger 2 metamorphosis in the colonial hydrozoan Hydractinia echinata

4 Huijuan Guo,1# Maja Rischer,1# Martin Westermann,2 Christine Beemelmanns1,*

5 1 Leibniz Institute for Natural Product Research and Infection Biology – Hans Knöll Institute, 6 Beutenbergstraße 11a, D-07745 Jena, Germany

7 2 Electron Microscopy Centre, Friedrich Schiller University Jena, Ziegelmühlenweg 1, D-07743 8 Jena, Germany

9 # contributed equally

10 * Email: [email protected] (ORCID 0000-0002-9747-3423)

12 Classification

13 Major: Chemical Sciences

14 Minor: Ecology

15 Keywords

16 Hydractinia, Pseudoalteromonas, metamorphosis, phospholipids, polysaccharides

17 Author Contributions

18 Conceptualization: M.R, H.G, M.W., C.B.; Methodology: M.R, H.G, M.W., C.B.; Investigations: 19 M.R, H.G, M.W., C.B.; Resources: M.W., C.B.; Writing – Original Draft: M.R, H.G, C.B. 20 Supervision and Funding Acquisition: M.W., C.B.

21 This PDF file includes:

22 Main Text 23 Figures 1 to 6 24

1 Abstract

2 Bacterial-induced metamorphosis of larvae is a widespread cross-kingdom communication 3 phenomenon within the marine environment and critical for the persistence of many invertebrate 4 populations. However, the chemical structures of the majority of inducing bacterial signals and the 5 underlying cellular mechanisms remain enigmatic. Hydractinia echinata larvae transform upon 6 detection of bacterial biofilm components into the colonial adult stage. Despite serving as cell 7 biological model system for decades, the inducing bacterial signals remained undiscovered. 8 Using a chemical-ecology driven analysis, we herein identified that specific bacterial 9 (lyso)phospholipids and polysaccharides, naturally present in bacterial biofilms, elicit 10 metamorphosis in Hydractinia larvae. While (lyso)phospholipids (e.g. 16:0LPG/18:1LPE, 16:0 11 LPA/18:1LPE) as single compounds or in combinations induced up to 50% of all larvae to 12 transform within 48 h, two structurally distinct polysaccharides, the newly identified Rha-Man 13 polysaccharide from Pseudoalteromonas sp. P1-9 and curdlan from Alcaligenes faecalis caused 14 up to 75% of all larvae to transform within 24 h. We also found combinations of 15 (lyso)phospholipids and curdlan induced the transformation in almost all larvae within 24 h, 16 thereby exceeding the morphogenic activity observed for single compounds and axenic bacterial 17 biofilms. By using fluorescence-labeled bacterial phospholipids, we demonstrated their 18 incorporation into the larval membranes, where interactions with internal signaling cascades 19 could occur. Our results demonstrate that multiple and structurally distinct bacterial-derived 20 metabolites converge to induce high transformation rates of Hydractinia larvae, which might 21 ensure optimal habitat selection despite the general widespread occurrence of both compound 22 classes.

24 Significance Statement 25 Bacterial biofilms profoundly influence the recruitment and settlement of marine invertebrates, 26 critical steps for diverse marine processes such as coral reef formation, marine fisheries and the 27 fouling of submerged surfaces. Yet, the complex composition of biofilms often makes it 28 challenging to characterize the individual signals and regulatory mechanisms. Developing 29 tractable model systems to characterize these ancient co-evolved interactions is the key to 30 understand fundamental processes in evolutionary biology. Here, we characterized for the first 31 time two types of bacterial signaling molecules that induce the morphogenic transition and 32 analyzed their abundance and combinatorial activity. This study highlights the crucial role of the 33 converging activity of multiple bacterial signals in development-related cross-kingdom signaling. 34 35 36

1 Main Text 2 3 Introduction 4 The radical transformation (metamorphosis) of planula larvae into the adult stage is a critical step 5 in the life cycle of many marine species as it confers the propagation and persistence of the 6 population in the marine ecosystem.1 For more than 80 years it has been recognized that 7 chemical signals present within marine bacterial biofilms induce or even prevent settlement and 8 metamorphosis in benthic marine larvae,2-4 but their identification remains still a challenging task 9 due to low production levels and unestablished model systems. Hence, until today only very few 10 key bacterial signals have been structurally characterized.5-7 A prime example represents the 11 bacterial product thallusin isolated from Zobellia uliginosa, which induces metamorphosis in the 12 alga Monostroma oxyspermum.8 Several members of the Polychaeta class and Cnidaria phylum 13 have also served as model systems for bacterial-induced metamorphosis over decades.9,10 In 14 several studies, it was found that bromopyrroles produced by Pseudoalteromonas induce larvae 15 of several coral species to undergo metamorphosis; however induced larvae failed to attach to 16 surfaces when stimulated by bromopyrroles alone indicating that other, yet unidentified, chemical 17 cues might be important for the morphogenic process.11,12 Recent biochemical investigations of 18 the bacteria-induced metamorphosis of the marine polychaete Hydroides elegans resulted in the 19 identification of a phage tail-like contractile injection systems (tailocins) in Pseudoalteromonas 20 species that induce settlement and metamorphosis by releasing an effector protein Mif1, which 21 stimulates the P38 and MAPK signaling pathways.13-15 However, bacteria not capable of 22 producing theses proteinaceous injection systems were also found to induce the transformation 23 releasing additional, yet structurally not defined morphogens.16,17 In the 1970s, Leitz and Wagner 24 reported that a lipid-like molecule of Pseudoalteromonas espejiana (original name: Alteromonas 25 espejiana) induces larvae transformation in Hydractinia echinata, an early branching metazoan 26 lineage dating back more than 500 million years.18,19 But despite intensive studies, the bacterial 27 signals causing Hydractinia larvae to metamorphose have remained elusive. Instead, 28 metamorphosis of Hydractinia was artificially induced using high salt concentrations (CsCl) 29 allowing seminal studies on migratory stem cells, allorecognition (self-recognition), the canonical 30 Wnt-signalling system, and the development of muscles and nervous systems.20,21 However, it 31 was noted from early on that artificial induction caused phenotypical and developmental 32 differences in Hydractinia development compared to bacterial induction.22 33 34 The long-standing unsolved question about the structures of bacterial signals and the apparent 35 morphological differences in larvae development between artificial and natural induction attracted 36 our interest. Thus, we set out to solve the structures of the bacterial signals that induce 37 metamorphosis in Hydractinia, which would allow us to shed light on the biochemistry underlying 38 this ancient prokaryote-eukaryote signaling mechanisms.23 39

1 Results 2 Bioassay-guided identification of bacterial signals 3 To investigate which co-occurring bacteria induce metamorphosis in H. echinata (hereafter 4 Hydractinia) we isolated morphological distinct co-occurring bacterial species from the surface of 5 a healthy and freshly collected Hydractinia colony.24 For a mono-species biofilm-based 6 metamorphosis assay, we selected 29 representative bacterial isolates, including seven genome- 7 sequenced Hydractinia-associated strains, one coral associated strain Pseudoalteromonas sp. 8 PS5,11,12 and eight bacterial type strains obtained from culture collections. Similar to previous 9 observations, we observed the inconsistent timing of metamorphosis using biofilms compared to 10 the artificial control (> 6 mM CsCl final concentration), presumably due the inhomogeneous 11 nature of biofilms and spatial concentrations differences of the yet unidentified inducing signals in 12 bacterial biofilms. To enable a comparative analysis of the inducing activity, we adopted the 13 established stage-chart of morphological appearances by Leitz and coworkers to describe and 14 calculate biofilm-induced rates of transformation (Figure S1).18,19 In short, the percentage of 15 larvae settlement includes only counts of larvae in morphological stages 9-11, and the 16 percentage of complete metamorphosis includes only counts of larvae in morphological stages 17 12-14. Based on these assessments, transformation rates are given as averaged percentage 18 ranges in 10% step ranges, which are calculated from replicates of experiments performed with 19 different larvae batches (different spawning events to ensure reproducibility of results). 20 As depicted in Figure 1, our bioassays resulted in the identification of five Hydractinia-associated 21 bacterial strains that caused up to 60% of all larvae to undergo complete transformation to the 22 primary polyp. Most notably, three strains (P1-9, P6-7, P1-29) rapidly induced the transformation 23 reaching the morphological stages 13-14 within 24 h. Additional five strains were found to induce 24 the transformation to stages 9-11 within the first 24 h, but subsequent development into the 25 primary polyp was only observed within 48 h. In addition, we observed that up to eleven strains 26 induced the transformation of larvae to morphogenic stage 9-11. However, neither of those 27 strains induced the full transformation to the primary polyp leading eventually to the death of the 28 transforming organism. In addition, six strains were found to be non-inducing and four strains 29 caused the death of up to 100% of all larvae within 24 h. The results of this study suggested firstly 30 that more than one structurally distinct bacterial signal might induce the transition and secondly 31 that the mode of action might differ depending on the structure of secreted molecule. Amongst all 32 tested strains, Pseudoalteromonas sp. P1-9 (referred from now on as P1-9) was found to induce 33 the most robust morphogenic response and thus was selected for further chemical analysis 34 (Figure 2).25-27 35 36 37 38 39 4

1 2 Figure 1. Assessment of morphogenic activity of selected bacterial strains. A) Left: phylogenetic 3 tree based on 16S rRNA gene sequences of 29 tested bacterial strains. Best DNA model was 4 generated and the robustness of interfered tree topologies was evaluated after 1000 bootstraps 5 (> 50% are shown). Right: Heatmap depicts transformation percentage (stage 9-11 and 12-14) of 6 H. echinata larvae counted after 24 and 48 h (negative control: ASW; positive control: CsCl). 7 Transformation rates are given as averaged percentage ranges in 10% step ranges, which are 8 calculated from replicates of experiments performed with different larvae batches. B) Hydractinia 9 echinata colonizing the shell inhabited by a hermit crab (Pagurus sp.). C-E) Microscopic image of 10 C) planula larva; D) initiation of morphogenic transformation by disc formation (morphological 11 stage 9-10); E) formation of primary polyp (morphological stage 14). 12 13 To test if the morphogenic cue of P1-9 is a secreted and/or a diffusible small metabolite, we first 14 tested solid-phase extracts (C18 cartridges) derived from liquid and plate cultures of P1-9. 15 However, none of the tested solid-phase extracts showed morphogenic activities compared to 16 living bacterial biofilms (positive control) (entry 3, Figure 2). We then tested if the signal is a

1 secreted high-molecular weight (HMW) biomolecule (e.g. protein, exopolysaccharides (EPS)) 2 and/or part of the bacterial membranes. 3 Size exclusion based separation of culture supernatants and cell membrane fragments resulted in 4 the isolation of a HMW fraction (> 30 kDa) that showed remarkably high morphogenic activity in a 5 dose-response manner (entry 5). In contrast, low-molecular weight fractions (< 5 kDa) showed 6 only moderate to very low morphogenic activities (entry 6). 7

8 9 Figure 2. Morphogenesis assay of biosamples derived from P1-9. Transformation rates to 10 morphological stages 12-14 were calculated from replicates of experiments performed with 11 different larvae batches after 48 h (‒ (no induction); + (< 40%; low induction); ++ (> 40% high 12 induction). 13 14 Enrichment of outer membrane vesicles and mini cells from culture supernatants by 15 ultracentrifugation induced high transformation rates already within the first 24 h resulting in fully 16 transformed larvae after 48 h (entry 7). Thus, we reasoned that the morphogenic cues might be 17 part of the outer cell membrane and/or of high molecular weight. To determine the stability of the 18 morphogens and to optimize purification methods, we subjected the active HMW fractions (> 30 6

1 kDa) to enzymatic and physical treatments and tested the morphogenic response after treatment. 2 As depicted in Figure 2, activity was completely retained when samples were treated with 3 digestive enzymes such as DNase, RNase, or proteinase K, or even when heated to 96 °C for 4 10 min (entries 9-12). Treatment of the active fraction with aqueous 6 M NaOH or 6 M HCl (12 h, 5 30 °C) partially solubilized the active morphogen (entry 13, 14), and activity of both, soluble 6 fraction and residue, was mostly retained after neutralization. However, treatment with 6 M HCl at 7 higher temperatures (> 60 °C) gradually abolished the morphogenic activity of the sample. Based 8 on these tests, we deduced that the morphogen(s) was neither a sensitive protein, nucleic acid 9 nor an instable secreted metabolite. Morphogens were also extractable from the bacterial 10 membranes and methanolic extracts showed indeed the highest morphogenic activity (entry 17, 11 18), while aqueous extracts retained only moderate activity (entry 15). Taken together, our results 12 indicated that presumably two different bacterial morphogens are produced by P1-9, which are 13 stable and extractable and likely localized within or at the bacterial outer membrane. 14 15 Analysis of morphogenic phospholipids 16 In a next step, we sought to characterize the morphogenic cues and applied a bioassay-guided 17 reverse-phase column chromatography (HPLC) purification protocol. The resulting active HPLC 18 fractions were analyzed by high-resolution tandem mass spectrometry (HR-MS2) and Global 19 Natural Product Social Molecular Networking (GNPS)28 analysis revealing the dominant presence 20 of different phosphatidylglycerol (PG) and phosphatidylethanolamine (PE) species and the 21 respective lyso derivatives (LPG, LPE). Due to inherent difficulties associated with the purification 22 of structurally closely related phospholipids, we purchased eleven commercial derivatives that 23 matched the proposed derivatives to compare the LC-HRMS2 pattern (Figure S12-S17), as well 24 as 1H and 13C nuclear magnetic resonance (NMR) spectra (chemical shifts and coupling 25 constants) (S28-S52).

26 27 Figure 3. A) 1H-NMR spectra of commercial 16:0-18:1 PG (red) and morphogenic PG/LPG 28 enriched fraction (blue). B-C) LC-HRMS2 based GNPS analyses of purified lipid fraction showing 29 the B) LPE and C) LPG MS2-cluster. 30 31 7

1 We then confirmed the morphogenic activity of commercial phospholipids including two 2 derivatives containing fluorescence labels (Figure 4) using 25 and 50 µM as final lipid 3 concentrations (based on preliminary dose-response tests using lipid extracts and commercial 4 16:0 LPG (Figure S15)). Here, it is noteworthy that phospholipid concentrations of single lipids or 5 combination thereof exceeding 50 µM occasionally caused the lysis of the organism (Figure 4). 6 Transformation rates were determined after 24 and 48 h, respectively; however statistical 7 analyses of assay data was not performed at this stage as induction by phospholipids resulted in 8 a non-synchronized behavior of larval transformation, similar to the observation made in axenic 9 biofilm assays. However, none of the tested single phospholipids recapitulated the full activity 10 observed for enriched phospholipid fractions derived from P1-9, but derivatives 16:0 LPG, 18:0 11 PE, and 16:0 (lyso)phosphatidic acid (16:0 LPA and 16:0 PA) repeatedly induced metamorphosis 12 in moderate to good rates (20-40%). 13 At this stage we questioned if the low solubility of (lyso)phospholipids might prevent the uptake 14 and perception by Hydractinia. Thus, we treated competent larvae with nitrobenzoxadiazole 15 (NBD)-labeled (phospho)lipids (18:1-12:0 NBD PG, 18:1-12:0 NBD PE or 12-NBD stearate 16 (Table S3), 25 nmol/mL) and found that all lipids were incorporated into the larvae membrane 17 within five hours and both phospholipid derivatives. While 12-NBD stearate did not induce 18 transformation to the primary polyp, 18:1-12:0 NBD PG and PE were found similarly active as 19 unmodified phospholipids (Figure 4). Here, it was noted that during metamorphosis the 20 fluorescence signal continuously decreased over time, presumably due to internalization and 21 decomposition of phospholipids during transformation processes (Figure 4, Table S3). Thus, it 22 was concluded at this stage that induction of metamorphosis by bacterial-derived phospholipids 23 presumably occurs by passive uptake of lipids into the larvae membrane, is strongly structure- 24 and concentration dependent and presumably orchestrated by Hydractinia-internal feedback- 25 loops. 26 Phospholipids (PL) and lysophospholipids (lyso-PLs, LPLs) are ubiquitous components of 27 bacterial cell walls and determine e.g. cell integrity and morphology as well as pathogen-host 28 interactions.29,30 However, little details are known about their possible signaling functions in 29 bacteria.31 As bacterial membranes and outer membrane vesicles (OMVs) are composed of 30 species-specific mixtures of (lyso)phospholipids, we hypothesized that the natural phospholipid 31 compositions of cells and bacterial biofilms will dictate the morphogenic activity and that additive 32 or even synergistic effects may play an important role in the induction process. 33

1 2 Figure 4. Metamorphosis inducing activity of commercial phospholipids. Transformation rates are 3 given as averaged percentage ranges in 10% step (calculated from replicates of experiments 4 performed with different larvae batches). Negative control: ASW; positive control: CsCl. A) 5 Bioactivity test were performed using 25 and 50 µM final lipid concentrations. Transformation 6 rates to morphological stage 9-11 and stage 12-14 were determined after 24 and 48 h. B-C) 7 Activity tests of phospholipid combinations (25 µM of each lipid in a 1:1 combination with a final 8 lipid concentration of 50 µM). Transformation rates to morphological stages 9-11 (B) and stages 9 12-14 (C) were determined after 48 h. D-K) Fluorescence labeling images of larvae treated with 10 (D, E) 18:1-12:0-NBD PE; (F, G) 18:1-12:0-NBD PA; (H-I) 18:1-12:0 NBD PG undergoing 11 metamorphosis (fluorescence imaging with emission wavelength at 527 nm). 12 13 Thus, we tested enriched phospholipid extracts (125 µg/mL lipid extract) of 16 different 14 Hydractinia-associated bacterial strains including strains that were found inactive in biofilm 15 assays for morphogenic activity (Figure S13). Indeed, extracts of several different strains (P1-23, 16 SW13, P1-9, P. rubra, P3-9, P1-4-1a, P1-16) induced the transformation to the primary polyp 17 (stage 12-14) in moderate to good rates within 48 h, in part exceeding the results of our previous 18 biofilm experiments (Figure 1, S13). In contrast, lipid extracts of four strains caused cell lysis 19 and/or abnormal transformations at the tested concentration, presumably due to the presence of 20 overall toxic phospholipid concentrations or known cytotoxic (hydrophobic) natural products 9

1 (violacein and/or poly-brominated pyrrole derivatives) naturally present within the bacterial cells 2 and membranes of the producing organisms. We also analyzed the phospholipid composition of 3 cell extracts by comparative high resolution tandem mass spectrometry (HRMS2) and found a 4 species-specific abundance of phospholipids that only in part resembled the composition of our 5 model strain P1-9. Despite the absence of detectable amounts of (L)PA derivatives, we 6 hypothesized that the combination of several weakly active (L)PE and (L)PG might be 7 responsible for the observed clear activity. To corroborate these findings, we then conducted 8 combination assays of a defined subset of phospholipids (50 µM final concentrations). Indeed, 9 several tested phospholipid combinations showed additive or even synergistic tendencies to 10 induce metamorphosis within 48 h, in particular the combinations of 18:1 LPE/16:0 LPG, 18:1 11 LPE/18:1 LPG, 16:0 PA/16:0 PG, 16:0 LPA/16:0 LPG and 16:0 LPA/18:1 LPE (Figure 4). At this 12 stage we concluded that specific (lyso) phospholipids and combinations thereof, known to be 13 present in bacterial cell membranes and OMVs, are in part responsible for metamorphosis of 14 Hydractinia larvae. 15 16 Analysis of morphogenic polysaccharide 17 As depicted in Figure 2, our bioassay-guided analysis indicated the presence of a second type of 18 morphogenic compound of high-molecular weight (HMW, > 10kDa in size) and naturally present 19 within aqueous extracts of Pseudoalteromonas P1-9 biofilms (Figure S9). Thus, we enriched for 20 the secreted bioactive compound by size-exclusion filtration and analyzed the most active 21 fractions by 1H NMR, which revealed a complex mixture of yet unknown polysaccharides

22 (Figure S21). Further purification (Sephadex G25, eluted with 0.1% NH4OH) and semi- 23 preparative HPLC (Shodex) and NMR and HRMS analysis of the most active fraction (> induction 24 of 20%) revealed the presence of a polysaccharide consisting of repeating -(1'→4)-α-L-Rha- 25 (1→3')-D-Man- units (> 10 kDa, Figure 5, Figure S19-S27). The rhamnose/mannose composition 26 was confirmed by acid hydrolysis using 6 M HCl, followed by TMS derivatization and GC-MS 27 analysis and comparison with commercial standards (Figure S18). The purified polysaccharide, 28 from now on named Rha-Man, showed a clear dose-dependent morphogenic induction of up to 29 80% transformation within 48 h (> 150 µg/mL, Figure S19). Partial acid hydrolysis resulted in loss 30 of morphogenic activity. At this stage, we also tested rhamnose and glucose monomers at 31 different concentrations, but no induction was observed. Comparative NMR-analysis of bioactive 32 size-exclusion fractions revealed that Rha-Man was only a minor component of the highly 33 bioactive polysaccharide fraction, which pointed towards additional not-yet structurally 34 characterized P1-9-specific exopolysaccharides (EPS) derivatives with morphogenic properties 35 (Figure S21).

1 2 3 4 Figure 5. Metamorphosis inducing activity of enriched and purified polysaccharides from P1-9 5 and commercial polysaccharides. Transformation rates are given as averaged percentage ranges 6 in 10% step (calculated from replicates of experiments performed with different larvae batches). 7 Negative control: ASW; positive control: CsCl. A) Bioactivity test were performed using (25 8 µg/mL) and transformation rates to morphological stage 9-11 and stage 12-14 were determined 9 after 24 and 48 h, respectively. B-C) Bioactivity tests of phospholipids (10 µg per lipid) and 10 curdlan combinations and 15 µg curdlan) were assessed after 24 h. D) Structures of repeating 11 units of morphogenic polysaccharides. 12 13 Due to the inherent difficulties associated with the structural characterization of EPS in general 14 and from P1-9 in particular, we decided to examine structure-function relations using structurally 15 defined and more widely distributed bacterial polysaccharides (charged and non-charged). 16 Intriguingly, curdlan, a well-known polysaccharide (50-200 kDa) with ß-1,3-glycosidic linkage and 17 produced by the Gram-negative bacterium Alcaligenes faecalis, induced at significant lower 18 dosage (25 µg/mL) the transformation to the primary polyp within 24 h (Figure 5). In contrast, 19 paramylon, the 500 kDa derivative of curdlan, which differs in its average length and three- 20 dimensional structure,32 induced the transformation only in moderate to low percentage. Other 21 EPS derivatives, in particular charged EPS (hyaluronic acid and heparin), induced only in very 22 low rates or were even toxic to larvae (Figure 5); secondary effects due to acidic/ionic properties 23 cannot be excluded for these derivatives. At this step, we concluded that Hydractinia larvae 24 respond in a dose-dependent fashion to the presence of two structurally distinct EPS derivatives 25 and fully developed primary polyp (> 50% transformation rates) within 24-48 h.

1 We then tested if phospholipids and the identified polysaccharides induced morphogenesis in an 2 additive or synergistic manner as both compound types co-occur within the bacterial biofilm. 3 Thus, we tested combinations of moderately active lysophospholipids, 16:0 LPG and 18:1 LPE, 4 as well as curdlan using lower concentration of individual compounds (10 and 15 µg/mL, 5 respectively). As expected, curdlan showed dose-dependent activity and induced transformation 6 to the primary polyp at lower rates compared to 25 and 50 µg/mL. Intriguingly, combinations of 7 curdlan with either 16:0 LPG, or 18:1 LPE, or both derivatives resulted in almost 100% 8 transformation rates within 24 h, which indicated towards synergistic (more than additive) 9 activities. In addition, we also observed that larvae triggered by the combination of natural 10 inducers developed more rapidly and cleanly into a fully functional primary polyp compared to 11 artificial induction.

12 13 14 Figure 6. Specific phospholipids, naturally present in bacterial cells and outer membrane 15 vesicles, and exopolysaccharides induce metamorphosis in H. echinata. A) Components of 16 bacterial biofilms recruit and induce metamorphosis of competent H. echinata larvae; or prevent 17 the transformation by killing larvae by secretion of toxic compounds. B) Scanning electron 18 microscopy of single cells of P1-9 obtained from a three day old liquid culture. C-D) Negative 19 contrast electron microscopic image of vesicles coated with S-layer like matrix and string-shaped 20 biopolymers isolated from a three day old liquid culture of P1-9 (C) and agar plate derived P1-9 21 biofilm (D). 22

1 The lipophilic nature of phospholipids and insolubility of polysaccharides raised the question of 2 how the bacterial signaling molecules are sensed and perceived by Hydractinia. In recent years 3 evidence has been mounting that extracellular (or outer membrane) vesicles (OMVs) represent a 4 highly conserved feature and potential mechanisms/vehicles of lipophilic signaling molecules that 5 orchestrate the interaction between bacteria and marine eukaryotes.33-35 To assess if P1-9 6 produces OMVs, we performed electron microscopy (TEM) imaging of shaking and static cell 7 cultures. Examinations of TEM images revealed entire cells and cells proliferating what appears 8 to be outer membrane vesicles (100-300 nm, Figure 6). Interestingly, (cryo)TEM imaging of 9 aqueous extracts of P1-9 biofilms and liquid culture revealed a high abundance of numerous 10 extracellular vesicles of different sizes and forms (Figure 6A,C,D) as well as high abundances of 11 biopolymer fibers (Figure 6B,C), which were positively tested for high morphogenic induction 12 (Figure 2). Thus, it can be concluded at this stage that morphogens are naturally present within 13 the biofilm matrix and are likely detected in form of OMVs and/or bacterial cell envelops. 14 15 Discussion 16 We successfully identified two types of bacterial morphogenic cues, a subset of 17 (lyso)phospholipids and two bacterial polysaccharides (Rha-Man, curdlan) from gram-negative 18 bacteria that induce the morphogenic transformation of Hydractinia planula larva into the primary 19 polyp. Detailed structure-activity studies of both compound types revealed a strong concentration 20 and structure-dependency of the morphogenic response. In case of phospholipids, 18:0 PE, 16:0 21 LPG and 16:0 LPA/PA showed the highest induction levels as single compounds and in 22 combination with other lipids, while other phospholipid combinations were found to be inactive or 23 even toxic to the larvae. Subsequent bioassays of phospholipid-rich bacterial extracts further 24 indicated that the overall compositions dictates the final morphogenic response and effects could 25 be overruled by the presence of toxic metabolites leading to abnormal or incomplete 26 transformation or even death of the larvae. In case of polysaccharides, a P1-9 derived Rha-Man 27 polysaccharide and A. faecalis-derived curdlan were found to induce the full transformation in a 28 dose-response fashion with curdlan being the more active derivative. Intriguingly, combinations of 29 phospholipids and curdlan induced the full morphogenic transition at rates exceeding in part the 30 sum of single compound contributions and resulting in the complete transformation of almost all 31 larvae within 24 h. Although, only representative derivatives of both bacterial compound classes 32 were investigated in this study, the results are of crucial importance for future research directions 33 in all facets for the following reasons: 34 Firstly, the structure, concentration and combination-dependency of the morphogenic response 35 hints towards species-specific cross-kingdom interactions that might allow for the optimal habitat 36 selection, despite the general global abundance of both compound classes in the bacterial 37 world.36-38 Here, it can be speculated that only specific bacterial biofilm compositions, which 38 induce the full transformation within less than 24 h, presumably enhance the likelihood of survival 39 of the individual larvae in highly competitive environments. Thus, our study will guide future 13

1 investigations into the beneficial microbiome and the long-term survival of naturally-induced 2 Hydractinia colonies. 3 Secondly, our studies pave the way to examine additive and synergistic behavior of different 4 bacterial signals that trigger animal signaling systems. Only few studies have so far focused on 5 synergistic effects of morphogens in the marine environment,4-7 which includes studies on 6 sulfonolipids (RIFs) and LPEs, that in combination induce the formation of predatory rosette-like 7 stage of the choanoflagellate S. rosetta,39-41 and the recent identification of synergistically acting 8 nucleobases from marine bacteria, that induce metamorphosis of the invasive fouling mussel 9 Mytilopsis sallei.42 However, detailed mechanistic studies about mode of perception and 10 synergism in most model systems are thoroughly missing. 11 Thirdly, the identification of a novel polysaccharide and the detection of other morphogenic, yet 12 uncharacterized derivatives, encourages research directed towards the elucidation of their 13 biosynthetic pathways and regulation and might enable their future structure elucidation and 14 biosynthetic engineering.43-45 15 Fourthly, for Hydractinia it has been proposed from early on that neurosensory cells dominantly 16 located at or near the anterior pole and in part on its tapered posterior tip are responsible for the 17 detection of the bacterial cues.19,20 Upon physical contact with lipid-rich vesicles and cell 18 fragments naturally present within marine biofilms, phospholipids could passively integrate into 19 the Hydractinia membrane; a hypothesis that is supported by our findings that fluorescence- 20 labeled phospholipids are quickly integrated into membranes of Hydractinia larvae. Once 21 integrated, specific phospholipids could act as ligands for certain receptors or induce changes in 22 membrane fluidity resulting in the recruitment of, e.g., PKC involved in cellular signaling 23 processes.46-48 In addition, LPAs have been recognized as potent mitogen in humans since 24 decades due to their interactions with G-protein-coupled receptors (GPCRs), thereby altering 25 many different cellular responses in humans, such as proliferation, survival, cytoskeletal changes 26 or calcium influx.49 Thus, a homologous mode of action could be hypothesized for Hydractinia. In 27 contrast, morphogenic EPS molecules presumably require detection via dedicated receptors as 28 described for EPS-mediated host-pathogen interactions.50,51 In particular curdlan has been 29 previously reported to act on lectin-type receptors in humans; a receptor-type, which have also 30 been detected in a Hydractinia transcriptome study.52 31 In summary, the results of our studies will allow us to establish Hydractinia as a model system to 32 investigate more closely host-microbe interactions and will guide future research on molecule- 33 receptor interactions. Our results also open several avenues for future studies on other aspects of 34 how bacterial signals trigger animal development in the marine world, which could have potential 35 practical applications for preventing biofouling, coral reef management and aquaculture 36 husbandry. 37 38

1 Materials and Methods 2 Supplementary Information (SI) available: details on fermentation, cultivation, bioassays, isolation 3 procedures, ESI-HRMS, 1H NMR, 13C NMR, and 2D NMR spectra as well as additional assay 4 data. 5 6 Acknowledgments 7 We are grateful for financial support from the German Research Foundation (DFG, CRC1127 8 ChemBioSys, BE 4799/2-1). CB greatly acknowledges funding by the ERC (ERC Grant number: 9 802736, MORPHEUS). MR is generously supported by the Jena School for Microbial 10 Communication (JSMC, DFG). Mrs. Heike Heinecke (HKI) for recording NMR spectra, Mrs. 11 Andrea Perner (HKI) for HRMS measurements, Toni Neuwirth for GC-MS measurements and 12 Johan Kufs for assisting in fluorescence imaging. We also would like to thank Prof. Valerie Paul 13 for the generous supply of Pseudoaltermonas sp. P5, and Theresa Jautzus for critical comments 14 on the manuscript. We are very grateful for the support by the Alfred-Wegner Sylt team and Ms 15 Birgit Hussel for their hospitality during field work. 16 17 Conflict of interest statement 18 The authors declare no conflict of interest. 19

20 21 22

1 References

2 1. M.-G. Hadfield, V.-J. Paul, “Natural chemical cues for settlement and metamorphosis of 3 marine invertebrate larvae” in Marine Chemical Ecology, J.-B. McClintock, B.-J. Baker, 4 Eds. (CRC Press LLC, 2001), pp. 431-461. 5 2. C.-E. Zobell, E.-C. Allen, The Significance of Marine Bacteria in the Fouling of 6 Submerged Surfaces. J. Bacteriol. 29, 239-251 (1935). 7 3. A.-M. Cantley, J. Clardy, Animals in a bacterial world: opportunities for chemical ecology. 8 Nat. Prod. Rep. 32, 888–892 (2015). 9 4. T. Wichard, C. Beemelmanns, Role of chemical mediators in aquatic interactions across 10 the prokaryote eukaryote boundary. J. Chem. Ecol. 44, 1008-1021 (2018). 11 5. M.-G. Hadfield, Biofilms and marine invertebrate larvae: what bacteria produce that 12 larvae use to choose settlement sites. Annu. Rev. Mar. Sci. 3, 453-470 (2011). 13 6. A. Woznica, N. King, Lessons from simple marine models on the bacterial regulation of 14 eukaryotic development. Curr. Opin. Microbiol. 43, 108-116 (2018). 15 7. M. McFall-Ngai, M.-G. Hadfield, T.-C. Bosch, H.-V. Carey, T. Domazet-Lošo, A.-E. 16 Douglas, N. Dubilier, G. Eberl, T. Fukami, S.-F. Gilbert, U. Hentschel, N. King, S. 17 Kjelleberg, A.-H. Knoll, N. Kremer, S.-K. Mazmanian, J.-L. Metcalf, K. Nealson, N.-E. 18 Pierce, J.-F. Rawls, A. Reid, E.-G. Ruby, M. Rumpho, J.-G. Sanders, D. Tautz, J.-J. 19 Wernegreen, Animals in a bacterial world, a new imperative for the life sciences. Proc. 20 Natl. Acad. Sci. USA. 110, 3229–3236 (2013). 21 8. Y. Matsuo, H. Imagawa, M. Nishizawa, Y. Shizuri, Isolation of an algal morphogenesis 22 inducer from a marine bacterium. Science. 307, 1598 (2005). 23 9. J.-D. Zardus, B.-T. Nedved, Y. Huang, C. Tran, M.-G. Hadfield, Microbial biofilms 24 facilitate adhesion in biofouling invertebrates. Biol. Bull. 214, 91-98 (2008). 25 10. S. Dobretsov, R.-M. Abed, M. Teplitski, Inhibition of biofouling by marine microorganisms. 26 Biofouling. 29, 423-441 (2013). 27 11. J.-M. Sneed, K.-H. Sharp, K.-B. Ritchie, V.-J. Paul, The chemical cue tetrabromopyrrole 28 from a biofilm bacterium induces settlement of multiple Caribbean corals. Proc Biol Sci. 29 281, 20133086 (2014). 30 12. J. Tebben, D.-M. Tapiolas, C.-A. Motti, D. Abrego, A.-P. Negri, L.-L. Blackall, P.-D. 31 Steinberg, T. Harder, Induction of larval metamorphosis of the coral Acropora millepora 32 by tetrabromopyrrole isolated from a Pseudoalteromonas bacterium. PLoS One. 29, 6, 33 e19082 (2011). 34 13. N.-J. Shikuma, M. Pilhofer, G.-L. Weiss, M.-G. Hadfield, G.-J. Jensen, D.-K. Newman, 35 Marine tubeworm metamorphosis induced by arrays of bacterial phage tail-like structures. 36 Science 343, 529-533 (2014). 37 14. N.-J. Shikuma, I. Antoshechkin, J.-M. Medeiros, M. Pilhofer, D.-K. Newman, Stepwise 38 metamorphosis of the tubeworm Hydroides elegans is mediated by a bacterial inducer 39 and MAPK signaling. Proc Natl Acad Sci U S A. 113, 10097-10102 (2016). 40 15. C.-F. Ericson, F. Eisenstein, J.-M.Medeiros, K.-E. Malter, G.-S. Cavalcanti, R.-W. Zeller, 41 D.-K. Newman, M. Pilhofer, N.-J. Shikuma, A contractile injection system stimulates 42 tubeworm metamorphosis by translocating a proteinaceous effector. eLife. 8, e46845 43 (2019). 44 16. M. L Freckelton, B. T. Nedved, Y-S. Cai, S. Cao, H. Turano, R. A. Alegado, M. G. 45 Hadfield; bioRxiv preprint first posted online Nov. 29, 2019; doi: 46 http://dx.doi.org/10.1101/851519 47 17. M.-L. Freckelton, B.-T. Nedved, M.-G. Hadfield, (2017). Induction of invertebrate larval 48 settlement; different bacteria, different mechanisms? Sci. Rep. 7, 42557 (2017). 49 18. T. Leitz, T. Wagner, The marine bacterium Alteromonas espejiana induces 50 metamorphosis of the hydroid Hydractinia echinata. Mar. Biol. 115, 173-178 (1993). 51 19. S. Seipp, J. Schmich, T. Kehrwald, T. Leitz, Metamorphosis of Hydractinia echinata— 52 natural versus artificial induction and developmental plasticity. Dev. Genes. Evol. 217, 53 385-394 (2007). 16

1 20. U. Technau, R.-E. Steele, Evolutionary crossroads in developmental biology: Cnidaria. 2 Development, 138, 1447–1458 (2011). 3 21. M. Steven, Sanders, Z. Ma, J. M. Hughes, B. M. Riscoe, G. A. Gibson, A. M. Watson, H. 4 Flici, U. Frank, C. E. Schnitzler, A. D. Baxevanis, M. L. Nicotra, CRISPR/Cas9-mediated 5 gene knocking in the hydroid Hydractinia symbiolongicarpus. BMC Genomics. 19, 649 6 (2018). 7 22. M. Kroiher, S. Berking, On natural metamorphosis inducers of the cnidarians Hydractinia 8 echinata (Hydrozoa) and Aurelia aurita (Scyphozoa). Helgol. Mar. Res. 53, 118-121 9 (1999). 10 23. T. Schneider, T. Leitz, Protein kinase C in hydrozoans: involvement in metamorphosis of 11 Hydractinia and in pattern formation of Hydra. Roux's arch. dev. biol., 203, 422-428 12 (1994). 13 24. H. Guo, M. Rischer, M. Sperfeld, C. Weigel, K.-D. Menzel, J. Clardy, C. Beemelmanns, 14 Natural products and morphogenic activity of γ-Proteobacteria associated with the marine 15 hydroid polyp Hydractinia echinata. Bioorg. Med. Chem. 25, 6088-6097 (2017). 16 25. J.-L. Klassen, T. Wolf, M. Rischer, H. Guo, E. Shelest, J. Clardy, C. Beemelmanns, Draft 17 Genome Sequences of Six Pseudoalteromonas sp. Strains P1-7a, P1-9, P1-13-1a, P1- 18 16-1b, P1-25 and P1-26, which Induce Larval Settlement and Metamorphosis in 19 Hydractinia echinata. Genome Announc. 3, e01477-15 (2015). 20 26. M. Rischer, J.-L. Klassen, T. Wolf, H. Guo, E. Shelest, J. Clardy, C. Beemelmanns,Draft 21 Genome Sequence of Shewanella sp. P1-14-1, a Bacterial Inducer of Settlement and 22 Morphogenesis in Larvae of the Marine Hydroid Hydractinia Echinata. Genome Announc 23 4(1), e00003-16 (2016). 24 27. J.-L. Klassen, M. Rischer, T. Wolf, H. Guo, E. Shelest, J. Clardy, C. Beemelmanns, 25 Genome Sequences of Three Pseudoalteromonas Strains P1-8, P1-11 and P1-30 26 Isolated from the Marine Hydroid Hydractinia echinata. Genome Announc. 3, e01380-15 27 (2015). 28 28. M. Wang, J. J. Carver, V. V. Phelan, L. M. Sanchez, N. Garg, Y. Peng, D. D. Nguyen et 29 al. "Sharing and community curation of mass spectrometry data with Global Natural 30 Products Social Molecular Networking." Nature Biotechnology 34, 8, 828 (2016). 31 29. H. Sunshine, M. L. Iruela-Arispe, Membrane lipids and cell signaling, Curr. Opin. Lipidol. 32 28, 408–413 (2017) 33 30. L. Zheng, Y. Lin, S. Lu, J. Zhang, M. Bogdanov, Biogenesis, Transport and Remodeling 34 of Lysophospholipids in Gram-negative Bacteria. Biochim. Biophys. Acta. Mol. Cell. Biol. 35 Lipids. 862, 1404-1413 (2017). 36 31. K. Makide, H. Kitamura, Y. Sato, M. Okutani, J. Aoki, Emerging lysophospholipid 37 mediators, lysophosphatidylserine, lysophosphatidylthreonine, 38 lysophosphatidylethanolamine and lysophosphatidylglycerol. Prostag. Oth. Lipid M. 89, 39 135-139 (2009). 40 32. R. Zhang, K. J. Edgar, Properties, chemistry, and applications of the bioactive 41 polysaccharide curdlan, Biomacromolecules 15, 1079–1096 (2014) 42 33. G. Raposo, W. Stoorvogel, Extracellular vesicles: exosomes, microvesicles, and friends. 43 J. Cell Biol. 200, 373–383 (2013). 44 34. S.-J. Biller, F. Schubotz, S.-E. Roggensack, A.-W. Thompson R.-E. Summons, S.-W. 45 Crisholm, Bacterial vesicles in marine ecosystems. Science. 343, 183–186 (2014). 46 35. J. Lynch, R.-A. Alegado, Spheres of hope, orbs of doom: the good and bad of outer 47 membrane vesicles in interspecies dynamics, J. Bacteril. 199, e00012-17 (2017). 48 36. E. Maunders, M. Welch, Matrix exopolysaccharides; the sticky side of biofilm formation. 49 FEMS Microbiol. Lett. 364, fnx120 (2017). 50 37. M.-B. Lilledahl, B.-T. Stokke, Novel imaging technologies for characterization of microbial 51 extracellular polysaccharides. Front. Microbiol. 6, 525 (2015). 52 38. D. Parkar, R. Jadhav, M. Pimpliskar, Marine bacterial extracellular polysaccharides. J. 53 Coast. Life. Med. 5, 29-35 (2007).

1 39. R.-A. Alegado, L.-W. Brown, S. Cao, R.-K. Dermenjian, R. Zuzow, S.-R. Fairclough, J. 2 Clardy, N. King, A bacterial sulfonolipid triggers multicellular development in the closest 3 living relatives of animals. eLife. 1, e00013 (2012). 4 40. C. Beemelmanns, A. Woznica, R.-A. Alegado, A.-M. Cantley, N. King, J. Clardy, 5 Synthesis of the rosette-inducing factor RIF-1 and analogs. J. Am. Chem. Soc. 136, 6 10210-10213 (2014). 7 41. A. Woznica, A.-M. Cantley, C. Beemelmanns, E. Freinkman, J. Clardy, N. King, Bacterial 8 lipids activate, synergize, and inhibit a developmental switch in choanoflagellates. Proc. 9 Natl. Acad. Sci. U S A. 12, 113, 7894-7899 (2016). 10 42. J. He, Q. Dai, Y. Qi, P. Su, M. Huang, C. Ke, D. Feng, Bacterial nucleobases 11 synergistically induce larval settlement and metamorphosis in the invasive mussel 12 Mytilopsis sallei. Appl. Environ. Microbiol. doi:10.1128/AEM.01039-19 (2019). 13 43. N.-A. Komandrova, M.-S. Kokoulin, A.-I. Kalinovskiy, S.-V. Tomshich, L.-A. Romanenko, 14 V.-E. Vaskovsky, The O-specific polysaccharide from the marine bacterium 15 Pseudoalteromonas agarivorans KMM 255T. Carbohydrate research, Carbohydr. Res. 16 414, 60-4 (2015). 17 44. D.-H. Limoli, C.-J. Jones, D.-J.Wozniak, Bacterial extracellular polysaccharides in biofilm 18 formation and function. Microbiol. Spectr. 3 (2014). 19 45. A. Casillo, R. Lanzetta, M. Parrilli, M. M. Corsaro, Exopolysaccharides from Marine and 20 Marine Extremophilic Bacteria: Structures, Properties, Ecological Roles and Applications. 21 Mar. Drugs 16, E69 (2018) 22 46. Y.-C. Yung, N.-C. Stoddard, H. Mirendil, J. Chun, Lysophosphatidic Acid signaling in the 23 nervous system. Neuron. 85, 669-682 (2015). 24 47. G. Amador-Cano, E. Carpizo-Ituarte, D. Cristino-Jorge D, Role of protein kinase C, G- 25 protein coupled receptors, and calcium flux during metamorphosis of the sea urchin 26 Strongylocentrotus purpuratus. Biol Bull. 210, 121-131 (2006). 27 48. G. Henningi, D.-K. Hofmann, Y.-B. Yahu, Metamorphic processes in the soft corals 28 Heteroxenia fuscescens and Xenia umbellata: The effect of protein kinase C activators 29 and inhibitors. Invertebr. Reprod. Dev. 34, 35-45 (1998). 30 49. M.-E. Lin, D. R. Herr, J. Chun, Lysophosphatidic acid (LPA) receptors: signaling 31 properties and disease relevance. Prostaglandins Other Lipid Mediat. 91, 130–138 32 (2010). 33 50. P. Albersheim, A.-G. Darvill, M. McNeil, B.-S. Valent, J.-K. Sharp, E.-A. Nothnagel, K.-R. 34 Davis, “Oligosaccharins: Naturally occurring carbohydrates with biological regulatory 35 functions”, in Structure and Function of Plant Genomes, O. Ciferri, L. Durelll, Eds 36 (Springer, 1983), pp. 293-312. 37 51. U. Szewzyk, C. Holmström, M. Wrangstadh, Relevance of the exopolysaccharide of 38 marine Pseudomonas sp. S9 for the attachment of Ciona intestinalis larvae. Ecol. 39 Progress. Ser. 259-265 (1991). 40 52. J. Soza‐Ried, A. Hotz‐Wagenblatt, K.‐H. Glatting, C. del Val, K. Fellenberg, H. R. Bode, 41 U. Frank, J. D. Hoheisel, M. Frohme, The transcriptome of the colonial marine hydroid 42 Hydractinia echinata. FEBS J 277, 197-209 (2009).